毕业项目：检测分神司机¶
In [3]:
%config InlineBackend.figure_format = 'retina'
from urllib.request import urlretrieve
from os.path import isfile, isdir, join, pardir
from IPython.display import SVG, Image
import os, shutil
import random
import glob
import zipfile
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
#import pydot
import cv2
import h5py
一、数据探索¶
driver_img_list.csv列出了训练集中的司机id，类别id和图像文件名称。10个类别id如下：
- c0: 安全驾驶
- c1: 右手打字
- c2: 右手打电话
- c3: 左手打字
- c4: 左手打电话
- c5: 调收音机
- c6: 喝饮料
- c7: 拿后面的东西
- c8: 整理头发和化妆
- c9: 和其他乘客说话
查看每个类别包含的图像数量。
In [4]:
driver_image_list = pd.read_csv(join(pardir, 'data', 'driver_imgs_list.csv'))
sns.set(style="darkgrid")
ax = sns.countplot(x='classname', data=driver_image_list, color='blue')
每个司机拥有的图像数量。
In [15]:
ax = sns.countplot(y='subject', data=driver_image_list, color='blue')
每个司机在每个类别上的图像数量分布。
In [16]:
ax = sns.factorplot(x="classname", col='subject', col_wrap=6, data=driver_image_list, kind="count", size=2.5, color='blue')
二、数据预处理¶
解压图像压缩包
In [17]:
driver_dataset_folder_path = join(pardir, 'data', 'driver_dataset')
zip_path = join(pardir, 'data', 'imgs.zip')
# 解压图像数据压缩包
if not isdir(driver_dataset_folder_path):
with zipfile.ZipFile(zip_path) as zipfile:
zipfile.extractall(driver_dataset_folder_path)
zipfile.close()
查看单张图像文件。
In [5]:
image_train_folder_path = join(pardir, 'data', 'driver_dataset', 'train')
image_test_floder_path = join(pardir, 'data', 'driver_dataset', 'test')
# 示例图片
image_file = join(image_train_folder_path, 'c0', 'img_34.jpg')
image = cv2.imread(image_file)
image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
print("The size of the image is {}:".format(image.shape))
plt.axis('off')
plt.title("c0: safe drive")
plt.imshow(image)
plt.show()
In [13]:
def show_img(classname):
img = glob.glob(join(image_train_folder_path, classname, "*.jpg"))
plt.figure(figsize=(16, 16))
for i in range(15):
plt.subplot(3, 5, i+1)
image_idx = random.randint(0, 100)
image = cv2.imread(img[image_idx])
image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
plt.axis('off')
plt.imshow(image)
In [14]:
show_img("c0")
In [68]:
show_img("c1")
In [69]:
show_img("c2")
In [70]:
show_img("c3")
In [71]:
show_img("c4")
